The O flaw

The explanation of how the O flaw came about has recently come to the fore due to it's inclusion of Mr Granzow's book. I have since seen a presentation that at least partially backs up Granzow's interpretation. Granzow starts his explanation with Bacon, as per below.

As we have said, only one of the impressions on a roller was usually used in laying down a plate, but during the first year or so it is possible that more than one roller impression was employed in a few instances. This, at any rate, seems to have been the course followed in the case of Plates7, 8, 9and 10 of the One Penny value. Most of the stamps on these four plates were produced from a defective roller impression, or impressions, and show a small flaw between the letters"O"and "N"of the word"ONE," which appears white on the stamps. This flaw probably arose from a speck of dirt or metal getting on to the original engraved die, or the surface of the roller, unnoticed, at the time the roller used for these plates was made. From the fact that a certain number of the stamps do not show this particular flaw it appears as if more than one roller impression was used in the making of these plates. As the flaw is found on four plates it was probably on the die, as it indicates that more than one impression on the roller received it before it was removed. We are unable to give the exact date when the roller with the defective impression or impressions was constructed, as the Engraving Book of the firm contains no entries of the rollers for the plates for many years, but it was almost certainly made during the first half of the month of May, 1840.' Bacon Vol 1 p56.

And:

Mr. Bertram McGowan, in an article published in 1918… As regards Plates 8, 9 and 10 he has found specimens with the same lettering on each of these plates with and without the flaw between the letters"ON" of the word"ONE." We have not seen the specimens included in his list, but we are not convinced that the presence or absence of the flaw is sufficient to prove that the plates were repaired by re-entering these impressions. We think that the difference was either caused in the process of printing, the disappearance of the flaw being accounted for by an extra application of ink and not such a careful wiping of the plates as usual, or that it was due to the wear of these particular impressions on the plates. Possibly both these causes, acting together, were contributing agents in the occasional disappearance of the flaw. Bacon Vol 1 p78-9.

Continuing from this Mr Granzow states:

The only way for a white flaw to have been produced from a particle on the roller were if it were pressed on or below the roller surface. It is thus more likely that it was picked up from the die. If the flaw were very thin, it could have disappeared early in the use of the plate. The position is strengthened by the fact that some positions appear free of the flaw in later printings while some do not. Figure 140 is a scan of the BB position from plate 10. The stamp on the left is an early printing showing the flaw between the 'O' and the 'N' in the bottom lettering. The stamp on the right is from a later printing of the same position showing some wear. The flake of steel in the early state of the plate was worn or broken off in use and the 'O' flaw disappeared. Granzow p205.

Lets analyse this.

“The only way for a white flaw to have been produced from a particle on the roller were if it were pressed on or below the roller surface.”

I am not sure of the logic here. In order for a piece of metal to create a white flaw, it would have been necessary for that particle to stop the roller from filling a depression on the die. The effect would be that an area that would have otherwise been raised on the roller was in fact flat or even a depression. If this roller had have been used on a plate, it would not have made a depression that would have held ink, and may even have left a raised area of metal that would have needed rubbing flat before printing started. Perhaps this is what Granzow meant by his statement.

“If the flaw were very thin, it could have disappeared early in the use of the plate... The flake of steel in the early state of the plate was worn or broken off in use and the 'O' flaw disappeared.”

I feel that this shows a lack of understanding of the process at work here. Granzow seems to be suggesting that the roller somehow either imparts a flake of metal onto every position on the plate that shows the flaw, or manufactures a simple machine that is activated by wear of the plate. This is clearly impossible. It could be that I have misunderstood what he is trying to get at.

Lets take a step back and look at the problem. A key point is the fact that the flaw is a progressive one. It changes. So was the problem with the die, the roller or the plates?

Die?

Since the roller only visited the die once and yet the flaw gets larger as time progresses, that would discount the problem being with the die.

For metal on the die to have been the cause of the white flaw, the development of the 'O' flaw would have to be reversed. That is, stage 3 of the flaw would have been the first impression taken. As the transfer roller was rotated to the next impression, a part of the offending metal on the die fell off, making a smaller (stage 2) flaw. The next impression on the roller, and another fleck of metal gone would have produced stage 1. The problem that I have with this is that usage on the plates appear backwards, although it could explain why plates have a mixture of 'O' flaw stages. Also the flake of metal would have to have been harder than the roller to make an impression and yet softer than the die so that it did not damage it. Also, making absolutely sure that the die was clean would have been a fundamental step for the engravers to take. And, why with stage 2 and 3 is the flaw surrounded by a coloured area? I don't see how this can be correct.

Plate?

Could the problem lie within the impressions on the plates – nearly 1000 of them? Well, both Bacon and Granzow seems to think that wear may play a part in the appearance or otherwise of the flaw, however this would need an explanation of a simple mechanism to achieve this, something I cannot envisage. Wear of the plates means that there is less ink captured by the impression, not more. As wear progresses we should see the areas of white expanding, not contacting. Granzow's comment “The flake of steel in the early state of the plate was worn or broken off in use and the 'O' flaw disappeared.” is clearly wrong.

Roller?

On balance, I believe that the problem lies in the transfer roller itself. The transfer roller just after its manufacture was composed of soft steel. In this state it was rolled onto the die a number of times to produce a number of roller impressions on its circumference. Then the roller was hardened by heating up to a cementing heat (ideally around the 1050-1100 deg C mark) for a good few hours in a container stuffed with carbon, then quickly quenched. Steel is almost unique in becoming stronger by this mechanism, whereas most metals become weaker. This process tended to very slightly deform the metal, which is one of the main reasons why Perkins Bacon tried to use the same roller impression for repairs (by re-entry) as that that laid down the plate.

The main point to remember was that metallurgy at this point in time was in its infancy. One of the big problems was minimising impurities, two of the biggest culprits being sulphur and phosphorus. Here is what Michael Siverns [retired consultant metallurgist] had to say on the subject:

“Phosphorus forms iron phosphide which dissolves in ferrite greatly weakening it. The general limit for phosphorus <0.06% in carbon steels and even lower in alloy steels. Up to the advent of "Basic" steel making in the 1850's phosphorus was avoided by the use of Swedish iron extracted from very pure ores with no phosphorus and incidentally low sulphur content. The "Basic" refers to bases in chemistry are are non-acid and react with phosphorus to form phosphates which are removed with slag. The bases are present both as special refractories in the furnace lining and in the flux (slag making material). Perkins -Bacon's dies especially the early ones could have used Swedish iron in the crucible process or selected another proven iron of similar quality - at high financial cost.

Sulphur forms sulphides which occur as iron sulphide or manganese sulphide. Iron sulphide forms yellow-brown inter-crystalline films which dramatically reduce the impact resistance of steel. Manganese sulphide forms as dove-grey globules which become inclusions in the ingot; however, they are relatively ductile when at the high temperature at which rolling takes place and form elongated string or thread like inclusions. These are much less harmful to the tensile and shock resistance of the steel than iron sulphide films. Thus steel-makers aim to ensure any sulphur is in the form of manganese sulphide. To do this successfully the manganese content of the steel needs to be five time the theoretical amount required to combine with the total sulphur content. This puts a practical limit of <0.06%S.

Thus with the printing plate both Sulphur and Phosphorus would need to be less than 0.06% each.”

It should be obvious where I am going with this.

I believe that the transfer roller was contaminated at the point where the 'O' flaw occurs on the transfer impression. The roller went through the hardening process and the soft carbon steel became hard, whereas the contaminated alloy remained relatively brittle. The roller was then used to make the impressions starting with plate 7. All was well at the start, but some way through the process the thin sliver of metal that runs from the base of the O sheared off the roller, perhaps unnoticed. See the attached diagrams that show the incidence of the different stages of the flaw. It would seem that there was another small sliver that fell off when rolling in the H row of plate 7 (stage 1a?), as Gibbons states that stage 2 can be seen faintly on some impressions. Looking through the Nissan reconstruction on the Spink site https://www.spink.com/nissen-reconstructions.aspx I would suggest that there are stamps that show a similar weakness in this area further down the sheet. The engraver continued until another larger lump fell off the transfer roller, producing state 2. It is reasonable to assume that stage 2 produced a pip above the plate level, and therefore it would have become all too evident the moment the engraver run his hand over the work. The roller stabilised at this point until the last few impression on plate 10, when another chunk of metal broke away from the roller.

There are some observations that I think are worth making. State 1 of the flaw is sometimes very difficult to see. If the metal that broke away was just below the level of the plate, it would have been possible for a small amount of ink to remain and disguise the flaw. This may provide an explanation of the point that Bacon was making when he said “we are not convinced that the presence or absence of the flaw is sufficient to prove that the plates were repaired by re-entering these impressions.” Further wear of the plate would have made the flaw more obvious as the level of the plate would have dropped down to the level of the flaw. The point at which the flaw occurred had nothing to do with it's position relative to the roller level, but had everything to do with the position of the impurities.

There is also the peculiarity of the distribution of the flaws over the plates. Namely, why do we find both stages 1 and 2 on plates 7 and 8? The only way for this to have happened is if production of plate 7 was interrupted and the transfer roller was used to start the rolling in of plate 8. The only explanation that I can come up with is that the engraver became aware of the problem on plate 7 soon after the later stage one flaw occurred. This plate was removed from the press and passed over to someone else to find out the feasibility of correcting this flaw by hand re-touch. Since there was still some pressure to produce these plates quickly, the engraver got on with plate 8, despite the known problem, at which point (during the O row on plate 8?) another lump fell off the transfer roller. After much swearing, I guess that the plates were swapped over again, plate 7 was finished off at the transfer press, whilst the plate 8 started it's repair by hand.

I have always been uncomfortable with the fact that the O flaw never seemed to behave itself, in that it never looked like its drawing. If I am correct, this should not be surprising, since every position would have had to have been retouched by hand, before it went to press. This point also solves the question why a few units do not show any sign of the O flaw at all, especially on plate 10. If the plates were on the repair bench anyway, any impression that really was a mess would have merited extra attention from the bruin. I would suggest that lack of time precluded the total repair of all impressions, the main goal was to make sure that a perfectly flat surface were produced in order to facilitate proper ink wiping and make printing possible.

I have not mentioned it, but all of the above comments apply to the NE ray flaw that also occurs on these plates. Same probable cause and similar results.

There is some indirect evidence to back up the possibility of contamination. Granzow has conducted some useful research into the suppliers of steel plates at this time, see chapter 2 of his book. In 1840 Perkins Bacon purchased some steel from Richard Bayley of Sheffield, enough for 40 plates. It would seem that the quality of the steel used to produce these plates was not perhaps up to the standard promised in the sales pitch, it's probably safe to safe to say not up to Swedish iron quality. In 1841, there was a letter from Bayley stating that he was having problems finding steel of a good enough quality. In late 1840, Perkins Bacon started to use John H Winder for their steel supplies and stayed with them until the end of the contact to produce postage stamps. It follows that it was probably poor quality steel supplied by Bayley that goes to the root of the problem of the O flaw.

AP

May 2014

I have received some pertinent and useful comments from Arthur Hick who suggested that a possible reason for stage 1 of the flaw appearing on both plates 7 and 8 could have been because the engraving room was undertaking some night shift work. He also points out that it would have been unlikely for them to continue using the roller when they knew that it was producing pips all over the plate.

The engraving books records the plates finished (not necessarily hardened) on:

• Plate 7 – 19^th June
• Plate 8 – 31^st July
• Plate 9 – 7^th October
• Plate 10 – 2^nd December

So there was no need for night work on these. They were not a rush job.

As far as I know there was one chap that did the engraving (Mr AE Fisher) and another that struck the letters. I would imagine that there was also an apprentice who would have been employed using the burin, cleaning up the plates after the above operations. Did they let the apprentice loose on a night shift? It is possible, but I don't think they did. Whatever the reason, it would seem that they were working on plates 7 & 8 at the same time for part of the early engraving process. Arthur has suggested that a natural break may be the cause, such as the normal plate lock up in the evening.

My crude drawings show stage 2 of the flaw eating into the roller so that a pip was raised on the plate. If the amount of metal breaking from the roller was less dramatic, so that the plate produced a flat or only a very slightly raised area on the plate, it could have been that they were able to print from it without repairing the plate in any way. I think Arthur is right to be uncomfortable with the scale of the problem shown by my drawing. Also, there is the possibility that the damage was more progressive than two or three occurrences. Perhaps only small amounts of metal were consistently flaking off the roller rather than relatively large catastrophic events as described above. However, the difference between stage 2 and stage 3 is significant. Here I believe that a pip of metal was produced on the plate which would have needed attention. This is why, I believe there are some state 1 plate 10s that do not show any sign of the O flaw. They have been repaired by hand.

Many thanks to Arthur Hick.

AP

March 2016